[Bugme-new] [Bug 4191] New: Process hangs while running file system stress test on XFS

[bugme-daemon at osdl.org]

http://bugme.osdl.org/show_bug.cgi?id=4191

           Summary: Process hangs while running file system stress test on
                    XFS
    Kernel Version: 2.6.11-rc3-mm1
            Status: NEW
          Severity: normal
             Owner: xfs-masters at oss.sgi.com
         Submitter: sharyathi at in.ibm.com
                CC: sglass at us.ibm.com


Distribution: SLES 9 
Hardware Environment: x235, 2 way, memory 1.5 GB
Software Environment: 2.6.11-rc3-mm1
Problem Description:  We started stress tests on mounted XFS file system, after
5-6 hours we noticed rm command getting hung and we are not able to delete all
the files from the XFS file system and the file system couldn't be unmounted.
The test process can be killed but even after that the rm command called by the
test process still remained. 

The stack trace of the rm process is as follows:
rm            D E5851F94     0  1077  32083                     (NOTLB)
c76dbf38 00000082 c1de6540 e5851f94 f3c12580 f3c125ac e5851f94 e3d5ca60
       c0116efb 00000001 00000001 c1c11020 00000001 0003640e 27a1b000 000f9b95
       00000006 e3d5ca60 e3d5cbb4 00002000 00000001 f348b7cc f348b7d4 00000246
Call Trace:
 [<c0116efb>] do_page_fault+0x1ab/0x57a
 [<c03089eb>] __down+0x7b/0xf0
 [<c011a7c0>] default_wake_function+0x0/0x10
 [<c0169270>] filldir64+0x0/0x100
 [<c0308b87>] __down_failed+0x7/0xc
 [<c0169428>] .text.lock.readdir+0x8/0x20
 [<c01693d7>] sys_getdents64+0x67/0xb0
 [<c01685d5>] do_fcntl+0x115/0x170
 [<c0103e19>] sysenter_past_esp+0x52/0x79

Stack trace of xms deamons running at that instance is as follows:

xfslogd/0     S 00000002     0 12979     19         12980  1123 (L-TLB)
xfslogd/1     S C01EAC2A     0 12980     19         12981 12979 (L-TLB)
 [<f928df10>] pagebuf_iodone_work+0x0/0x40 [xfs]
xfslogd/2     S C01EAC2A     0 12981     19         12982 12980 (L-TLB)
 [<f928df10>] pagebuf_iodone_work+0x0/0x40 [xfs]
xfslogd/3     S C0119AD5     0 12982     19         12983 12981 (L-TLB)
xfsdatad/0    S C0123194     0 12983     19         12984 12982 (L-TLB)
xfsdatad/1    S C0119AD5     0 12984     19         12985 12983 (L-TLB)
xfsdatad/2    S C0119AD5     0 12985     19         12986 12984 (L-TLB)
xfsdatad/3    S C0119AD5     0 12986     19         20555 12985 (L-TLB)
xfsbufd       S C0104D9A     0 12987      1         12989 11221 (L-TLB)
 [<f928ec4a>] pagebuf_daemon+0x6a/0x1e0 [xfs]
 [<f928ebe0>] pagebuf_daemon+0x0/0x1e0 [xfs]
xfssyncd      S C1C0AF9C     0 12989      1         32410 12987 (L-TLB)
 [<f9293c8b>] xfssyncd+0x7b/0x1b0 [xfs]
 [<f9293c10>] xfssyncd+0x0/0x1b0 [xfs]
 [<f92734b0>] xlog_grant_push_ail+0x130/0x170 [xfs]
 [<f9269572>] .text.lock.xfs_iget+0x72/0x160 [xfs]
 [<f928775f>] xfs_lock_inodes+0xaf/0x120 [xfs]
 [<f928769a>] xfs_lock_dir_and_entry+0xda/0xf0 [xfs]
 [<f9287927>] xfs_remove+0x157/0x3f0 [xfs]
 [<f9280c00>] xfs_trans_unlocked_item+0x20/0x40 [xfs]
 [<f9280c00>] xfs_trans_unlocked_item+0x20/0x40 [xfs]
 [<f9285e97>] xfs_access+0x37/0x40 [xfs]
 [<f9291bb0>] linvfs_permission+0x0/0x20 [xfs]
 [<f9291913>] linvfs_unlink+0x13/0x40 [xfs]
 [<f926defc>] xfs_ichgtime+0xfc/0x104 [xfs]
 [<f9291bb0>] linvfs_permission+0x0/0x20 [xfs]


Steps to reproduce:
mount a xfs file system 
and run File System Stress test on the File System.
after running it for 5-6 hours we found out we were not able to unmount the file
system.
Note: We encountered this problem after following these steps, not sure if the
deffect can be reproduced.

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.